An illustration of one of the major rice producing areas of Malaysia, Sekinchan

What does Microsoft LUIS do and how it affects your AI Chatbot quality?

Enhance your Azure Bot with Microsoft LUIS

To improve our chatbot quality, we do not stop at a basic QNA Chatbot or a menu-driven chatbot. We need a technology to predict users’ overall intent and extract detailed information at the same time. This is when Microsoft Language Understanding (LUIS) comes into picture when building an AI Chatbot.

The portal for Microsoft Language Understanding (LUIS)

Microsoft LUIS is a cloud-based conversational AI service from Microsoft Azure which uses custom machine-learning intelligence. Basically, Microsoft LUIS allows your bots or applications to understand what a person wants in their own words. Yes, not only bots can use this cognitive service but other application as well since it has endpoints API. Some basic non-bot’s usage examples are extracting key information from a customer’s complaint email.

So, how did we use Microsoft LUIS to improve our AI Chatbot built with Microsoft Azure Bot Framework? Reads on to find out more.

Microsoft LUIS’s Intent

What is an intent? Intent is defined as something we plan or mean to do. Basically, it is the intention or purpose or a user when talking to the chatbot. The user might want to make an appointment, to find nearest petrol station or just want to greet the virtual assistant.

Microsoft LUIS makes it easy to detect the intent of a user. We just need to provide some user utterances and group them into intents. Utterances are inputs to the AI chatbot that need to be interpreted. To maintain a good chatbot, collect utterances that you think users will enter in the beginning and continue to monitor and train the chatbot. Over the time, the chatbot will understand the users more.

LUIS is used to detect user’s intent of looking for nearest petrol station.

An example of chatbot understands the user’s intent to find nearest petrol station is shown in the image above.

LUIS is used to detect user’s intent of submitting a complaint.

An example of virtual assistant understands the user’s intent to submit a complaint in the image above.

Microsoft LUIS’s Prebuilt Entities

Entities are another information that we can extract from user’s input via Microsoft LUIS. First type of entities is the prebuilt entities which is very easy to use. We just need to enable it via the portal. Then, we can detect the prebuilt entities. Some examples of the prebuilt entities are person name, date time and email address.

LUIS is used to extract name and phone number

In the example above, LUIS is used to extract person’s name, email, and phone number accurately from user’s input. This is important so that our virtual assistant can greet back the user using his name and confirmation of his phone number. If there is no entities detection, we can only greet the user with “Thanks, My name is Alex, our salesperson will contact you soon at my number is 012345678’, for example. This is because we have to assume the user’s reply is the name, email, and phone number.

The prebuilt datetime entities of LUIS is also very useful. If a user said next Monday, LUIS would return the exact date of next Monday. This makes the conversation more natural during an appointment making, for example.

There are many more prebuilt entities which you may refer to Microsoft’s documentation https://docs.microsoft.com/en-us/azure/cognitive-services/luis/luis-concept-prebuilt-model

You may ask, what if the entities you need are not available in the prebuilt entities? LUIS allows us to create our own entities. Most basic entity types would be list entity and regex entity. In list entities you can also provide synonyms. For example, an entity for small size can includes “sm, sml, tiny, smallest” as its synonyms.

Another example of entities would be BMW the brand entities which you can provide synonyms as “Bayerische Motoren Werke AG”, “BM”, “Bayerische Motoren Werke GmbH” and “Bavarian Engine Works Company”. As for regex entity, you can use it for fixed format such as identity card number.

There are many more ways you can enhance your chatbot with LUIS. Do check out Microsoft’s documentation at this link or check out some of our chatbots that are using Microsoft LUIS and Azure Bot Framework:

  1. MyMesra’s Meva
  2. InvestKL’ Kayla
  3. Affin Hwang Asset Management’s Nadia

XIMNET is a digital solutions provider with two decades of track records specialising in web application development, AI Chatbot and system integration.

XIMNET is launching a brand new way of building AI Chatbot with XYAN. Get in touch with us to find out more.

--

--

--

Digital Agency in Malaysia | www.ximnet.com.my

Love podcasts or audiobooks? Learn on the go with our new app.

Recommended from Medium

Stock Price Prediction using Machine Learning

GPT-3’s closest competitor!

Big Picture of Machine Learning

Paper repro: “Learning to Learn by Gradient Descent by Gradient Descent”

Machine Learning for Beginners

Deploy your own SSDLite Mobiledet object detector on Google Coral’s EdgeTPU using Tensorflow’s…

[ACM TELO 2021 / NeurIPS 2020 Works] Reusability and Transferability of Macro Actions for…

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
ximnet

ximnet

Digital Agency in Malaysia | www.ximnet.com.my

More from Medium

NLP-based book recommendation

Semantic search in risk management using NLP pipeline

ADI, a parameterizable and content-based recommender system — the fundamentals (Part 2)

Quick Tip: How to save spaCy dependency graphs as png